Thank you for sending your enquiry! One of our team members will contact you shortly.
Thank you for sending your booking! One of our team members will contact you shortly.
Course Outline
Insight into the Chinese AI GPU Landscape
- Comparative analysis of Huawei Ascend, Biren, and Cambricon MLU
- Distinguishing between CUDA and CANN, Biren SDK, and BANGPy paradigms
- Market trends and vendor ecosystem developments
Migration Preparation
- Evaluating your existing CUDA codebase
- Defining target platforms and SDK versions
- Setting up toolchains and development environments
Code Translation Methodologies
- Adapting CUDA memory access patterns and kernel logic
- Translating compute grid and thread models
- Exploring automated versus manual translation approaches
Platform-Specific Development
- Utilizing Huawei CANN operators and custom kernels
- Implementing the Biren SDK conversion workflow
- Reconstructing models using BANGPy (Cambricon)
Multi-Platform Testing and Tuning
- Profiling execution efficiency on each target platform
- Optimizing memory usage and comparing parallel execution
- Monitoring performance and iterative refinement
Managing Hybrid GPU Environments
- Deploying hybrid solutions with multiple architectures
- Developing fallback mechanisms and device detection strategies
- Implementing abstraction layers for sustainable code maintenance
Practical Case Studies and Best Practices
- Porting vision and NLP models to Ascend or Cambricon
- Adapting inference pipelines for Biren clusters
- Resolving version conflicts and API discrepancies
Conclusions and Future Directions
Requirements
- Practical experience in programming with CUDA or GPU-based applications
- A solid grasp of GPU memory models and compute kernel design
- Knowledge of AI model deployment or acceleration workflows
Target Audience
- GPU developers
- System architects
- Specialists in software porting
21 Hours
